There are 14 variables totally that need to be explored, they are:

Preform the same EDA based on each of them, finding if any of them could have significant effect on outcome.

For each test, as different patients will take different times of this test, we need to consider one, or more than one, uniform measurement(s) to measure the value of the test. Here, I will consider four measurements: mean, min, max, and changing pattern of test value over times. For the variable changing pattern, if the test value increases obviously, or said the sum of the difference between previous test score and next test score is larger than 2 units, over times, it will equal to 1; if the test value decreases obviously over times, it will equal to -1; if the test value is stable, then it will equal to 0.

GCS:

There are 4891 patients that have the GCS data, and one patient would take this GCS test for most 43 times.

Glucose:

There are 4798 patients that have the Glucose data, and one patient would take this Glucose test for most 18 times.

HCO3:

There are 4877 patients that have the HCO3 data, and one patient would take this HCO3 test for most 19 times.

HCT:

There are 4914 patients that have the HTC data, and one patient would take this HTC test for most 20 times.

HR:

There are 4891 patients that have the HR data, and one patient would take this HR test for most 152 times.

K:

There are 4839 patients that have the K data, and one patient would take this K test for most 18 times.

Also, as the test score of this K test are all less than 10, I change to view sum of difference between previous test score and next test score larger than 1 as increasing pattern, less than 1 as decreasing pattern.

Lactate:

There are 2471 patients that have the Lactate data, and one patient would take this Lactate test for most 20 times.

MAP:

There are 3423 patients that have the Glucose data, and one patient would take this Glucose test for most 154 times.

MechVent:

There are 3055 patients that have the Glucose data, and one patient would take this Glucose test for most 34 times.

Mg:

There are 4814 patients that have the Glucose data, and one patient would take this Glucose test for most 18 times.

Na:

There are 4876 patients that have the Glucose data, and one patient would take this Glucose test for most 23 times.

NIDiasABP:

There are 4308 patients that have the Glucose data, and one patient would take this Glucose test for most 97 times.

NIMAP:

There are 4300 patients that have the Glucose data, and one patient would take this Glucose test for most 95 times.

NISysABP:

There are 4312 patients that have the Glucose data, and one patient would take this Glucose test for most 98 times.